Impact of Health Insurance
on Access, Use, and Health
Status in Costa Rica
James Cercone, Etoile Pinder, Jose Pacheco Jimenez, and Rodrigo Briceno
Costa Rica, as a middle-income country that has largely achieved uni-
versal health coverage, allows for analysis of the diﬀerences in behavior
and care for the small share of citizens who remain uninsured. is
chapter sheds light on the impact of being covered by insurance in a
country where access is guaranteed even if uninsured and on the costs
and beneﬁts of covering the last 10%–20% of the population with
insurance or other approaches.
A country of 4.5 million people, Costa Rica has a per capita gross
domestic product (GDP) of US$5,600 (US$10,700 in purchasing power
parity terms), and in 2007 it spent about 7.1% of GDP on health care.
In 2008 the infant mortality rate was less than 10 deaths per 1,000 live
births, and average life expectancy was 80 years for women and 76 years
for men. Average life expectancy exceeds that of the United States by a
year, even though U.S. GDP per capita is four times that of Costa Rica.
Costa Rica has mandatory health insurance coverage, established
in 1941, and a comprehensive primary health care model that reaches
all citizens. e Caja Costarricense del Seguro Social (the Caja) is an
autonomous government institution that is both insurer and provider
of care. Nearly 90% of the country’s 4.5 million people are covered.
e health insurance system is based on traditional Bismarkian social
insurance, with an expanded role of the central government to cover the
uninsured population. It provides equal access to health care services,
irrespective of income or contribution. e formal sector contributes
90 Chapter 5
14.75% of payroll income to sustain the system. e poor and indigent are cov-
ered by the “noncontributory” and “insured by state” regimes, which have led to
equal access to health services for the poor and wealthy, something not seen in
any of Costa Rica’s neighboring countries. In addition, the absence of copayments
removes another possible barrier to equal access.
Main characteristics of the Costa Rican health system
e Costa Rican health system includes a wide range of entities; the most relevant
for this study are the Ministry of Health, the National Insurance Institute (INS),
and the Caja. e ministry oversees the performance of the essential public health
functions and exercises the stewardship role in the health sector, while the INS
oﬀers protection against occupational risks and traﬃc accidents as well as accident
liability and a voluntary insurance plan for health care.
e Caja is the key institution for this study. It manages and organizes manda-
tory health insurance and is an autonomous institution with technical, administra-
tive, and functional independence. It manages the compulsory health insurance
funds that come from payroll taxes and provides the highest proportion of health
care services in the country, covering roughly 90% of the population with a broad
package of services. Besides health services, it provides social security protection
to insured individuals and poor households through the Disability, Old Age, and
Funeral regime. In Costa Rica there is an administrative purchaser-provider split
between the ﬁnancial network and the provider network of the Caja. Nearly all
provision is through the Caja network; however, the Caja also contracts with pri-
vate providers and nongovernmental organizations for some services.
e network of providers belongs to the Caja, which is organized as a pyramid-
style network with primary care at the bottom and tertiary hospital care at the top.
Primary care consists of 104 health regions and 953 basic care teams (Equipos Bási-
cos de Atención Integral en Salud, or EBAIS). Each EBAIS covers 3,500–4,000
people and consists of a general practitioner, an auxiliary level nurse, and a pri-
mary care technician. All members must be registered with a primary care provider.
Recently, the Caja has expanded its purchasing options, and some primary care ser-
vices, such as minor surgeries and diagnoses, are purchased from nonpublic agents.
Secondary care consists of 10 major clinics, 13 suburban hospitals, and 7
regional facilities specializing in hospital services. Tertiary care has three general
hospitals and ﬁve specialized hospitals (women, children, geriatrics, psychiatry, and
rehabilitation). General and regional hospitals have a set number of people in their
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 91
catchment areas, so clients cannot freely select their hospital—that is, every person
must register in the hospital in the zone where he or she lives.
e health care sector also has a private subsector, which has expanded sub-
stantially in recent years, exempliﬁed by its increasing number of medical profes-
sionals. In the 1990s the share of medical staﬀ in the private sector increased from
10% to 24%. Household surveys show that about 30% of the population uses
private health services at least once a year.
Eligibility and coverage
According to the constitution and founding laws of the Caja, mandatory health
insurance in Costa Rica covers the risks of illness, maternity, disability, aging, and
involuntary unemployment. e insurance also partly covers ﬁnancial burdens due
to maternity, widowhood, orphanage, and burial. All wage-earners must be cov-
ered by health insurance, which also covers the workers’ dependents. Coverage
for poor families and the indigent is deﬁned by the Caja board of directors and
is ﬁnanced by the government from general tax revenues. All pensioners are auto-
matically covered by health insurance.
Health insurance was originally established to protect workers against the risk
of illness, maternity, and labor injuries, and initially no other group was covered by
the insurance. e mandatory health insurance scheme has evolved over time. In
1961 congress established universal health insurance for workers and their families.
e system expanded rapidly in the 1960s, when coverage almost tripled to half the
population. In 1975 health insurance was expanded to cover agricultural workers.
In 1978 the Caja created the voluntary health insurance scheme for independent
workers. In 1984 it created the special regime, funded from general tax revenues,
to cover the indigent. In 1993 all provision of care was moved to the Caja, and
the Ministry of Health became the regulator and coordinator of the sector. e
fundamental diﬀerence between Costa Rica and the rest of Latin America is that
Costa Rica uniﬁed its public delivery system under social security, eliminating the
parallel system operated by the Ministry of Health.
In the 2000s the approval of the Act for the Worker’s Protection set the manda-
tory enrollment of independent workers. Recent decisions by the Caja board have
aimed at expanding coverage to speciﬁc vulnerable groups. For instance, the ben-
eﬁts of health insurance can now cover a brother or sister of a contributory member
if he or she is disabled or taking care of their parents.
Roughly 88% of Costa Rica’s residents have health insurance. Coverage
increased from 60% in 1975 to a high of 92% in 1990. Since then, coverage rates
have remained at 86%–88%.
92 Chapter 5
e Caja law establishes universal coverage of health services—that is, no per-
son can be denied health care services even if uninsured. Emergency services are
free to both the insured and the uninsured. If uninsured persons go to a public
facility, two possibilities arise. First, they can pay for the services. Second, if they
lack resources and are deemed indigent, they can become insured by the state.
e process of aﬃliation and access to the health insurance system
Any person living in Costa Rica may become aﬃliated with the social health insur-
ance scheme in any one of at least ﬁve ways.
• As a formal or self-employed worker. Health insurance is mandatory in Costa
Rica for all categories of paid workers. Until 1999 this obligation was deﬁned
for formal salaried workers only. Since then, the Workers’ Protection Act estab-
lished that self-employed workers must also be aﬃliated with the Caja pro-
gram. Migrants are covered by the same legislation as citizens or residents. e
contribution rate for formal employees is a 14.75% payroll tax and is 10.25% of
reported income for independent workers.
• As a pensioner. All pensioners, either members of the contributory or noncon-
tributory schemes, are automatically aﬃliated with the national health insur-
ance program. Noncontributory pensioners receive a pension from the state even
though they never contributed. Contributions are set at 14% for pensioners.
• As state-covered members. Under this program, aﬃliation is usually deﬁned at
demand—that is, the uninsured person ﬁrst receives health care services at the
public facility, then is requested to pay for the services or to become enrolled
by the state, based on a means test applied by the Caja. e program is funded
by special taxes on luxury goods, liquor, beer, cola, and other similar imported
goods. It accounts for about 12% of the insured.
• As a voluntary member. A special option gives individuals the possibility to
enroll in the insurance program if they do not belong to any of the other
groups. In this option, the person enrolls and pays a regular fee of about $25 a
month. ere is no diﬀerence in the package of services this category receives.
• As an indirect member. Dependents of members are automatically covered.
Spouses and children studying until age 25 are covered by the direct member.
As mentioned above, some additional options exist, such as dependency status
for a contributing member’s brother or sister who is taking care of his or her
parents and is older than 60.
If a person cannot be included in any of the ﬁve groups, the chances of being
insured are essentially nil because of the absence of private health insurance options
prior to 2009. e Free Trade Agreement with the United States opened the Costa
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 93
Rican insurance market, a public monopoly since 1924, allowing international
insurance companies to establish, operate, and sell insurance plans.1 Currently, the
INS, the public insurance monopoly, operates one health insurance scheme with
approximately 10,000 aﬃliates nationwide.2 Some transnational companies have
insured their employees in the past through international health insurance pro-
grams, allowing their workers to obtain health care in private hospitals and clinics.
Table 5.1 summarizes sources of revenue for the Caja. e contributory portion of
the system accounts for about 76% of revenues. For poor households, the Caja actuar-
ial department estimates the number of poor households based on a household survey.
en it estimates an average premium, based on average wages and the application of
a 14% payroll tax rate to that average. e number of estimated poor people times the
average premium determines the expected global contribution from the state.
Who are the uninsured?
An analysis of the National Health Survey (ENSA 2006) shows signiﬁcant dif-
ferences in the sociodemographic proﬁles of the insured and uninsured in Costa
Rica (table 5.2; see box 5.1 for a description of the data and methodology used
in the analysis). According to the survey, 81% are insured and 19% are not. e
uninsured are less likely to be female (54% of the insured are female, compared
with 46% of the uninsured) and less likely to be married (32% of the insured are
married, compared with 20% of the uninsured). No signiﬁcant diﬀerence in aver-
age age exists between the two groups, yet the age structure is substantially diﬀer-
ent (ﬁgure 5.1). Adults ages 19–54 comprise 50% of the insured but 62% of the
uninsured. Adults ages 55 and older comprise 16% of the insured but 12% of the
Payroll fees by insurance scheme, 2006
Type of health insurance
Contribution by type of contributor (%)
Employee Employer State
regime Tot al
Salaried 5.50 9.25 0.25 —15.0 0
Independent 4.75 —5.50 —10.25
Voluntary 4.65 —5.50 —10.15
Contributor y pensioner 5.00 —0.25 8 .75 14.00
Noncontributor y pensioner — — 0.25 13 .75 14.00
Insured by stat e — — 14.0 0 —14.00
— not applicable.
Source: Caja Costarricense del Seguro Social.
94 Chapter 5
uninsured. ese statistics suggest that there can be self-selection of the healthy
into uninsured status even in a system with universal coverage.
e uninsured are more likely to have completed secondary education and to
work for a small company with fewer than 10 employees. ey are signiﬁcantly
more likely to be independent workers—17% of uninsured are self-employed, com-
pared with 9% of the insured—and to have a lower monthly income ($135 for
the un insured, compared with $159 for the insured). e uninsured are also more
likely to be immigrants (11% of the uninsured, compared with 3% of the insured).
Immigrants tend to work in low paying informal sector jobs, and their employers
Sociodemographic statistics for the insured and uninsured, 2006
N=3,98 8 (81.30%)
N=917 (18.70%) Difference
Age Year s 31.323 20. 310 30 .778 18.036 0.545
Sex Female 0.536 0.499 0.458 0.498 0.078 ***
Income Per capita
income (US$) 159.11 153.77 134.95 145.9 0 24.16 ***
1st quintile 27.41 16.78 26.30 18.16 1.11
2nd quintile 71.31 11 .31 73.13 11. 81 –1. 81
3rd quintile 115 .17 14.77 114.77 13.40 0.41
4th quintile 185.39 26.17 183.68 23.60 1 .71
5th quintile 401.86 190.82 378.27 225.25 23.59
Civil status Married 0.322 0.467 0.200 0.400 0.122 ***
Household head Household head 0.214 0.410 0.236 0.425 – 0.022
Education Primary 0.487 0.500 0.472 0.499 0.015
Secondary 0. 377 0.485 0.420 0.494 –0.043 **
University 0 .113 0 . 317 0.096 0.295 0 .0 17
Employment Self-employed 0.086 0.280 0.172 0.377 –0.086 ***
Size of rm Small rms have
(0=small, 1=big) 0 .073 0.260 0.0 21 0.14 5 0.052 ***
Nationality Costa Rican 0.967 0.17 9 0.894 0.308 0.073 ***
Urban status 1 = urban 0.684 0.465 0.662 0.473 0.022
*** signicant at p < 0.01; ** signicant at p < 0.05.
Source: ENSA 2006.
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 95
Data and methodology
We use data from three sources: the
nationally representative National Health
Survey (ENSA 2006), the administrative
database of hospital discharges for 2006
(Caja Costarricense del Seguro Social
2006), and the nationally representative
Income and Expenditure Survey (IES
2004). For ENSA there are a total of 7,522
households. We removed 54 with missing
insurance status as well as 1,409 public
employees and their dependents and 1,154
pensioners (because neither group has a
choice whether to be insured, and our
population of interest is those who have
a choice). For the hospital data we also
removed public employees and pensioners,
which reduced the number of discharges
from 326,583 to 267,325. We deleted
these observations because the only people
with the ability to avoid enrolling are the
self-employed and private employers who
choose illegally not to enroll themselves or
Our strategy for identifying the im-
pact of insurance was to ﬁnd instrumental
variables that would explain aﬃliation
with the Caja but not health or ﬁnancial
protection behavior. Depending on the
dataset, we used a subset of the follow-
ing for this purpose: size of ﬁrm where
the individual is employed (we used the
mean size within a household), sector of
economic activity, occupation type, self-
employment, having multiple jobs, and
canton code. No exclusion restrictions are
perfect, but we believe that in the context
of Costa Rica, these variables would af-
fect the dependent health variables only
through their impact on aﬃliation with
65 and older
65 and older
Age structure of the insured and uninsured, 2006
Source: ENSA 2006.
96 Chapter 5
often do not pay taxes for them. e authorities have few tools to detect and pre-
vent self-exclusion and to collect premiums for or from independent or itinerant
Insured people tend to have a higher burden of disease (a metric that summarizes
mortality and morbidity conditions in a determined population) and higher preva-
lence of speciﬁc diagnosed diseases, such as diabetes, hypertension, and asthma than
uninsured people (table 5.3). is does not imply a negative impact of insurance on
health status (a causal relationship cannot be inferred), but it can be considered that
if a person feels healthy, they are more inclined to believe that the beneﬁts of paying
insurance premiums outweigh the ﬁnancial cost. Of course, it is also consistent with
underdiagnosis in the uninsured population. But there is no statistically signiﬁcant
diﬀerence between the two groups’ self-perceived health status.
ere is no statistically signiﬁcant diﬀerence between the insured and unin-
sured for general utilization statistics (see table 5.3). For outpatient services, 65%
of the insured and the uninsured report having a visit during the last year. For hos-
pital services, 5% in both groups report a hospital admission for at least one night
in the last year. ere is only one signiﬁcant diﬀerence in use, but it is important:
while 49% of insured women ages 40 and older report having received a mammo-
gram, fewer than 40% of uninsured women report having received one. Given the
need for a patient to be referred to a high technology diagnostic imaging appoint-
ment, it is logical this would be an area where the uninsured are at a disadvantage.
According to the database of hospital discharges in 2006, 16% of people
discharged were uninsured, compared with 19% according to household survey
data (table 5.4). With childbirth as a leading cause of admissions, substantially
more women than men are in the hospital discharge database than in the ENSA
household data. But diﬀerences in insurance status are not large. ENSA shows that
85.3% of men and 79.7% of women are insured, while at discharge, 81.2% of men
and 85.4% of women were insured.
As with the survey data, the discharge data show that the uninsured were more
likely to be unmarried, but there is a signiﬁcant diﬀerence between the percentage
of survey respondents and discharges who declare they are “cohabitating.” Only
6% of insured survey respondents state they are cohabitating, while 19% of insured
discharges say they are. ere is an incentive to claim cohabitation with a Caja
member because it qualiﬁes the patient for dependent beneﬁts.
e discharge data paint a more nuanced picture of diﬀerences in behavior
between the insured and uninsured. e ﬁrst is how they are admitted. Some
87% of the uninsured were admitted to hospitals through the emergency room,
compared with 58% of the insured. While 39% of the insured are referred to the
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 97
Health status and use of insured and uninsured people, 2006
N=3,98 8 (81.30%)
N=917 (18.70%) Difference
More than one diag-
nosed disease 0.269 0.444 0.147 0.354 0.122 ***
Hypertension Diagnosed arterial
hypertension 0.095 0.293 0.039 0.195 0.056 ***
Diabetes Diagnosed diabetes 0.032 0.17 7 0.018 0.132 0 .014 **
Asthma Diagnosed asthma or
bronchitis 0.0 41 0.19 8 0.009 0.095 0.032 ***
Index, 0, lowest
burden of disease, to
100, highest burden
of disease 4.565 10.6 65 2.671 7. 955 1.894 ***
Scale, 1, ver y good,
to 5, very bad 2.085 0.707 2.124 0. 813 –0.039
Visit 1 = pers on visited
doctor at least once
during last year 0 .6 47 0.478 0.655 0.476 –0.009
Hospitalization 1 = per son hospital-
ized at least one night
in last year 0.052 0.2 21 0.050 0. 219 0.001
Emergency 1 = person used
at least once during
last year 0.117 0. 321 0.10 2 0.303 0.015
Mammogram Woman ages 40 and
older received mam-
mogram 0.490 0.500 0.397 0.4 91 0.094 **
Cytology Woman ages 18 and
older received pap
smear 0.934 0.248 0.921 0.271 0.014
Vaccines Person under age
vaccinations 0.826 0.379 0.812 0.392 0.015
diabete s medicines in
the two weeks pr ior to
the study 0.73 3 0.44 3 0.706 0.462 0.027
in the two weeks pr ior
to the st udy 0 .73 4 0.442 0.76 5 0.427 – 0. 031
*** signicant at p < 0.01; ** signicant at p < 0.05.
Source: ENSA 2006.
98 Chapter 5
hospital through an outpatient provider, only 8% of the uninsured follow this
route. What happens to them in the hospital is also diﬀerent. About 17% of the
insured and 21% of the uninsured have minor surgery, but 16% of the insured
undergo major ambulatory surgery, compared with only 2% of the uninsured.
e uninsured experience considerably longer stays (5.2 days, compared with 3.8
days) and are far more likely to end up in the intensive care unit (3% of uninsured
Comparison of insured and uninsured people who have been discharged
from a hospital, 2006
N=2 24,800 (84.1%)
N=42,525 (15.9%) Difference
Age Year s 28.48 18.567 28.23 19.662 0.253 **
Sex Female 0.700 0.458 0.634 0.482 0.066 ***
Civil status Married 0.358 0.479 0.186 0.389 0.172 ***
Nationality Costa Rican 0.930 0.254 0.771 0.420 0.15 9
Death Discharged dead 0.00 97 0.0 978 0.0296 0.170 0.0199 ***
3.75 8 5.026 5.238 6.951 –1.4 80 ***
admissions 0.047 0 .24 6 0.026 0 .17 7 0.022 ***
medical visits 1.339 2.892 1.062 2.456 0. 277 ***
0.387 0.487 0.082 0. 274 0.305
Emergency room 0.584 0.493 0. 8 74 0.332 –0.290 ***
Childbirth 0.029 0.16 8 0.044 0.206 –0.015 ***
0.098 0.298 0.136 0.343 –0.038 ***
Surgery 0.167 0. 373 0.209 0 .407 – 0.043 ***
Gyno-obstetrics 0.407 0.491 0.399 0.490 0.008 ***
Pediatrics 0.162 0.368 0.16 8 0. 373 –0.0 06 ***
surgery 0.15 6 0.363 0 . 017 0 .130 0 .139 ***
Psychiatry 0.008 0.087 0.040 0.197 –0.033 ***
Intensive care unit 0.002 0.048 0.029 0.16 7 –0.027 ***
*** signicant at p < 0.01; ** signicant at p < 0.05.
Source: Hospital discharges database.
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 99
admissions, compared with less than 0.3% of insured). e ENSA data show that
the insured appear to be in worse health status than the uninsured (only 14% of the
uninsured report having more than one diagnosed chronic condition, compared
with 29% of the insured). Finally, the uninsured have a statistically signiﬁcant
higher chance of inpatient death (3%, compared with only 1% for the insured).
In short, despite being healthier, the uninsured enter hospitals directly in more
apparent trauma and with less previous attention and planning than the insured.
ey are signiﬁcantly more likely to die. ese results have implications for patient
health outcomes and for health care costs, as hospitals are by far the most expensive
places to receive care.
What are the determinants of insurance status?
Probit analysis indicates that insurance status depends primarily on age, employ-
ment, nationality, education, and marital status (table 5.5). Not being married
signiﬁcantly decreases the chances of being insured (being married increases an
individual’s chance by 12 percentage points). Being an immigrant decreases the
likelihood of being insured by 19 percentage points relative to being a native Costa
Rican, and having less than a secondary education also decreases the probability.
Another large contributor is self-employment. Compared with being unem-
ployed (individuals employed by the government and pensioners are excluded from
the analysis because they have no choice whether to be covered), the likelihood of
being insured if an individual is self-employed is 12 percentage points lower, again
pointing to the independent worker’s decision to self-select out of the insurance
plan to avoid paying the tax—a problem in all payroll tax–ﬁnanced systems, made
worse as tax rates rise.
Impacts of health insurance
is section discusses the impact of insurance coverage on a number of outcome
variables, based on the estimation strategy shown in the appendix.
Does health insurance aﬀect access and use?
Using the ENSA 2006 data, we explored the impact of insurance for the full sam-
ple and for subsamples of the household data. ere is no evidence of a statistically
signiﬁcant diﬀerence between the insured and uninsured in the use of outpatient
care, hospitalizations, or emergency services. e lack of a diﬀerence extends to
insured and uninsured individuals from subsamples of the poorest 40% and the
wealthiest 40%. is result reinforces the descriptive data, which show few diﬀer-
ences between the insured and the uninsured.3
100 Chapter 5
We also analyzed subsamples with chronic conditions—one group diagnosed
with at least one disease, one group diagnosed with hypertension, and a third diag-
nosed with diabetes. ere is a statistically signiﬁcant impact only for diabetes:
health insurance reduces the probability of both inpatient care and emergency
room care for diabetics. Uninsured diabetics (all other things being equal) are more
likely to end up in an inpatient bed or the emergency room. In addition, insurance
reduces the use of medicines for the diabetic population. ese ﬁndings suggest
that insurance coverage in Costa Rica results in better and safer management of
diabetes, probably associated with primary care.
Probit analysis of the determinants of participation—dependent variable:
insurance status, 2006
Variable Partial ef fect Standard error Signicance
Age –0.004 0.002 *
Se x (wo men =1) 0.013 0.013
Log of the household monthly income per capita 0.018 0.008 **
Civil st atus (reference = single)
Married (married=1) 0.122 0.022 ***
Relationship to household head
Head of household (head of household=1) –0. 017 0.015
Education (reference = without formal education, primary incomplete and primary complete)
Secondary 0.027 0.015 *
University 0.033 0 .0 17 *
Employment category (reference = not employed)
Patron –0.077 0.062
Self-employed –0.125 0.060 **
Private 0 .0 17 0.0 41
Costa Rican (Costa Rican=1) 0.192 0.042 ***
Urban zone (urban=1) –0.007 0.013
Quality of health care services (self-reported) –0.022 0.006 ***
Reference population in the EBAIS to attend 0.000 0.000
Number of observations 3,070
Lod pseudo log likelihood –1,4 28 .41
Pr e d . P. 0.871
*** signicant at p < 0.01; ** signicant at p < 0.05; * signicant at p < 0.10.
Source: ENSA 2006.
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 101
Health insurance does not appear to have any impact on access to diagnostic tests
like pap smears in women ages 18 and older and mammograms in women ages 40
and older. e mammography result is diﬀerent from the descriptive analysis, indi-
cating that controlling for other factors (such as education) eliminated the disparity
between the insured and uninsured. In contrast, health insurance improves the likeli-
hood of completing the full series of vaccinations for children age 18 and younger.
From the 2006 hospital discharge data analysis, the insured were signiﬁcantly
less likely to access the hospital through the emergency room, consistent with the
descriptive data. e likelihood of having an avoidable hospitalization for a condi-
tion that could be managed in an outpatient setting was signiﬁcantly less for the
insured. is result, consistent with the household data result for diabetes, again
points to the importance of quality of health care services over quantity. Because
no one is denied care in Costa Rica, uninsured people who are sick are generally
able to receive treatment, but apparently often later than they should and in a less
than optimal setting. For health outcomes and cost-eﬀectiveness, the uninsured
are in a position inferior to the insured, who are enrolled with a primary care pro-
vider through their EBAIS, and therefore receive appropriate preventive and main-
tenance care. at, in turn, reduces chronic disease complications and expensive
hospitalizations for those diseases.
Does health insurance aﬀect health status?
ENSA includes the variable “self-reported health status,” which reﬂects whether
individuals describe their health as very good, good, okay, bad, or very bad. Over-
all, health insurance signiﬁcantly improves an individual’s self-perception of health
status. But it reduces the self-perceived health status for diabetics, interesting
because insured diabetics are less likely to need hospital services. Perhaps insured
diabetics are more educated about the serious complications associated with their
disease and thus consider themselves to be in a worse state of health.
From the hospital discharge data analysis, we have more objective information
on health status. Insured mothers are less likely to have babies with low birthweight,
which would be consistent with better access to prenatal care through EBAIS. In
addition, if we measure severity of illness by the number of days of hospitalization,
insured people experience substantially shorter hospital stays. When insured and
uninsured people are hospitalized, the insured are healthier by this measure.
Does health insurance aﬀect out-of-pocket expenditures?
e impact of health insurance on ﬁnancial protection of insured people was
estimated using a third data source, the nationally representative Income and
102 Chapter 5
Expenditure Survey of 2004. We estimated per capita out-of-pocket spending
on health as a proportion of per capita expenditures and as a proportion of pay-
ment capacity, deﬁned as total household expenditures minus household food
e average monthly out-of-pocket health expenditure by Costa Ricans in
2004 was US$8.50, but with a high degree of variability (coeﬃcient of variation
was estimated at 321%). Per capita out-of-pocket health expenditures represent
nearly 3% of per capita expenditure and 3.5% of the payment capacity of indi-
viduals. Out-of-pocket health expenditures represent only 2% of the poorest third
of the population’s per capita expenditures but 4% of the wealthiest third’s, the
reverse of the usual tendency in the absence of eﬀective ﬁnancial protection.
Our analysis found no signiﬁcant impact of health insurance on a Costa
Rican’s out-of-pocket health expenditures. It is likely this result is due largely to
the fact that no person can be denied care in Costa Rica. An individual who arrives
at an emergency room needing to be admitted to hospital is admitted, regardless
of the ability to pay. So there is no diﬀerence in out-of-pocket health expenditures
between insured and uninsured individuals.
e main distinguishing characteristic of health insurance in Costa Rica relative to
other countries of Latin America and many other middle-income countries is that
approximately 81% of the population is aﬃliated with the Caja and thus covered,
but even those who are not covered are guaranteed access to emergency and hospi-
tal care provided by the Caja when they are sick or need care. ey are not shunted
into a separate lower quality system. erefore everyone is covered by catastrophic
insurance; the major diﬀerence is that the 19% not aﬃliated with the Caja do not
beneﬁt from assignment to a primary care provider and must seek and pay for those
services in the market.
e uninsured are somewhat less educated and more likely to be immigrants,
have lower income, be self-employed, and come from healthier age groups than
the insured, but in their overall use of health care resources they are similar. If we
had only the household data, we would have concluded that the insured gained no
advantage over the uninsured except for a higher probability of children receiving
all immunizations and better care for diabetics.
However, the hospital discharge data raise concerns. e uninsured are far
more likely to enter the hospital through the emergency room. ey are likely to
have surgery but not to have a planned major surgery; even so, they experience a
36% longer stay. We estimate that simply reducing their length of stay to that of
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 103
the insured would save about US$8.5 million, or about US$100 per uninsured—
probably enough to ﬁnance a reasonable level of access to primary care for them.
e uninsured are more likely to be hospitalized for a condition that can be
managed in an ambulatory care setting, to end up in the intensive care unit, and
to be discharged dead. e hospital data analysis conﬁrms a higher probability of
emergency room and inpatient care for uninsured diabetics.
We expected to ﬁnd few diﬀerences between the insured and uninsured in
Costa Rica because of the equal access rule. For measures of ﬁnancial protection
and use of services, our expectations were met. Yet in signiﬁcant ways the unin-
sured are disadvantaged from a health standpoint. ey use medical care resources
more haphazardly than the insured. We hypothesize that this happens principally
because the insured enter the pyramidal Caja system at the bottom, or primary
care level, while the uninsured tend to enter closer to the top. Perhaps, in light of
these ﬁndings, it would be possible for the Caja to reallocate resources to cover the
uninsured in a more health-friendly manner.
104 Chapter 5
Appendix: Details of the Estimations
e estimated models when we use the ENSA 2006 database are the following:
Îi(insuredi) = a0 + a1sexi + a2agei + a3linci + a4marriedi + a5hohi + a6secondaryi + a7univi +
a8costaricani + a9employedi + a10zonei + Σ
15 arregionri + a16ivi + a17disease_statusi +
a18ebaisxhi + ei
Hi = b0 + b1sexi + b2agei + b3linci + b4marriedi + b5hohi + b6secondaryi +
b7univi + b8costaricani + b9employedi + b10zonei + Σ
15 brregionri + b16Îi(insuredi) +
b17disease_statusi + b18ebaisxhi + mi
where Hi is the result variable, sex is a dummy variable that takes the value 1 if the
individual is a woman and 0 if the individual is a man, age is the age of the indi-
vidual, linc is the logarithm of per capita household income, married is a dummy
variable that takes the value 1 if the individual is married and 0 otherwise, hoh is
a dummy variable that takes the value 1 if the individual is the head of the house-
hold and 0 otherwise, secondary (univ) is a dummy variable that takes the value 1 if
the individual has secondary (university) completed as his highest education level,
costarican is a dummy variable that takes the value 1 if the individual is a citizen
and 0 otherwise, employed is a dummy variable that takes the value 1 if the person
works and 0 otherwise, regionr are regional dummy variables, insured is a dummy
variable that indicates the coverage of the health insurance system (Îi(insuredi) is
the estimated value of this variable from the ﬁrst stage), disease_status is an index
that shows health status, and ebaisxh is the number of people in each health area.
For self-reported health status models we add other independent variables: consult
(a dummy variable that takes the value 1 if if the individual has visited a clinic
or hospital during the last year and 0 otherwise), hospital (a dummy variable that
takes the value 1 if the individual has been hospitalized at least one night during
the last year and 0 otherwise), and emergency (a dummy variable that takes the
value 1 if the person used emergency services at least once during the last year and
0 ot her wise).
Impac t of Healt h Insu rance o n Access , Use, and Healt h Status in C osta Ric a 105
e estimated models when we use the 2006 hospital discharge database are
detailed as follows:
Îi(insuredi) = a0 + a1sexi + a2agei + a3marriedi + a4costaricani + a5losi + a6previousi
13approvpi + a16ivi + ei
Hi = b0 + b1sexi + b2agei + b3marriedi + b4costaricani + b5losi + b6previousi + Σ
+ b14Îi(insuredi) + mi
where Hi is the result variable, sex is a dummy variable that takes the value 1 if the
individual is a woman and 0 if the individual is a man, age is the age of the individ-
ual, married is a dummy variable that takes the value 1 if the individual is married
and 0 otherwise, costarican is a dummy variable that takes the value 1 if the indi-
vidual is a citizen and 0 otherwise, los is the number of days an individual remained
in a hospital, previous indicates the number of earlier entrances to the clinic or
hospital, provp are province dummy variables, and insured is a dummy variable that
takes the value 1 if the individual is covered by the health insurance system and 0
otherwise (with Îi(insuredi) being the estimated value from the ﬁrst stage).
1. e market is to be opened in 2009 and 2012, depending on the type of insurance
2. e Free Trade Agreement breaks this monopoly so international insurance companies
can now arrive in Costa Rica, operate their oﬃces there, and oﬀer the public not only
health insurance but life and automobile insurance.
3. Insurance has a puzzlingly signiﬁcant negative impact on use of outpatient care for
non–Costa Ricans, but that is for a tiny subsample of only 175.